Ananish Chaudhuri: The trade off

There is a trade-off: how many lives will be taken by Covid-19 and how many lives will be lost due to our attempts to prevent loss of lives from Covid-19, writes Professor Ananish Chaudhuri of the University of Auckland

In his book “Risk Savvy”, the behavioural scientist Gerg Gigerenzer notes that, in the immediate aftermath of September 11, 2001, many Americans decided that flying was too risky. Instead, they chose to drive. In the 12 months following the attacks, an additional 1,500 people lost their lives on the road while trying to avoid the risk of flying. This is more than the total number of passengers in the planes used in the attack.

A similar phenomenon is playing out right now as the world essentially comes to a standstill to prevent deaths from Covid-19. But in doing so, we are focusing on what the psychologist Daniel Kahneman calls “identified lives”; the loss of lives that are right in front of us. Gigerenzer calls this the “fear of dread risk”: the apprehension about losing a lot of lives within a short time.

Every day, we learn how many people died of the coronavirus around the world. But many of those people would have died in the normal course of events, from a variety of reasons such as heart-attacks or flu. So, the issue is not so much how many people died of Covid-19, but how many more?

In focusing on identified lives, we ignore the loss of “statistical” lives. It is likely that the total impact of that loss will be greater than any loss of lives due to Covid-19. But those deaths will register less on our collective psyche since they will be diffused, scattered all over the world and will not be reported on in the same manner.

Like it or not, there is a trade-off here: how many lives will be taken by Covid-19 (identified lives) and how many lives will be lost due to our attempts to prevent loss of lives from Covid-19 (statistical lives).

In fact, at the time of writing, hospitals in Washington State, which has been hard hit by the virus, are engaged in a bleak triaging of which patients should receive treatment and which should not, since providing everyone with adequate treatment is no longer an option.

A recently released influential paper by Neil Ferguson and colleagues at Imperial College, London suggests: “Two fundamental strategies are possible: (a) mitigation, which focuses on slowing but not necessarily stopping epidemic spread and (b) suppression, which aims to reverse epidemic growth, reducing case numbers to low levels and maintaining that situation indefinitely.”

However, in reality these are part of a continuum: to what extent do we force people to self-isolate, curtail economic activity and reduce social contact? According to Ferguson and his colleagues, we will need to do this for 18 months.

But, bear in mind that we are really dealing with probabilities here. A shutdown of 18 months will work better than a shut-down of six months, which is better than four weeks; each imposing a different magnitude of social and economic costs.

Already, we are seeing a spike in unemployment claims and business insolvencies. We know that unemployment results in significantly lower life expectancy. It will also lead to homelessness and increased poverty. Mental health problems, particularly among children, will rise dramatically. All of these will also take a toll on healthcare systems and healthcare workers. The human cost of job losses and bankruptcies will be massive. Much of the pain of this shut-down will be borne by the socio-economically disadvantaged.

Beyond a certain point, it would just not be worth it to keep the economy shut down in order to save more people. Does the Government (or Treasury) have realistic estimates of how much the economy will shrink, how many jobs will be lost, how many businesses will go bankrupt? How large is the relief package required to prevent an economic catastrophe if the lockdown ends after four weeks or if it continues beyond that? Surely, this calculation should play a role and dictate how long a shut-down we can survive.

The real question here is: How large is the reproduction rate; i.e., the rate at which the contagion spreads? Current estimates suggest this rate for coronavirus is more than two; each infected person is affecting more than two others. We need to bring this number down as far as practicable.

It seems strict self-isolation should be able to achieve this goal. At this point, and I may well be proved wrong, Japan seems to be successful in keeping the spread down while allowing people to go about their business.

It is clear a crucial factor is population density. So a lockdown in places like Auckland or Wellington may make sense. It is not clear to me that large parts of the South Island, with low population density, need to be locked down.

For much of the country outside the large metropolitan areas, we should be able to do what we were doing before. Avoid large gatherings and implement self-isolation as needed. Complement this with public service announcements about good hygiene and the need to stay home if someone believes they may be infected in order to bolster the social norm of self-isolation.

Let people decide their risk-tolerances. Offer all those above 60, those with a history of respiratory problems or ones with compromised immunity the opportunity to work from home, should they choose to do so. Support businesses in providing paid sick leave to these workers over and above their usual entitlements.

What we face right now is a social dilemma; those who have been infected need to make sure that they do not spread the infection. But, evidence suggests Kiwis were and are doing a pretty good job with self-isolation. People who are obviously flouting self-isolation rules are being denied patronage at businesses and at times being reported to the police.

My research suggests people can be quite good at solving such collective action problems; that exhortative public messages asking people to choose cooperative actions can succeed. It may need to be backed up with sanctions for hard-core violators.

At the very least, the Government should track the path of the infection and selectively loosen restrictions in different parts of the country as and when appropriate. Ideally, much of the country should be restriction-free before four weeks have passed.

This will allow us to mitigate both the coronavirus catastrophe as well as the catastrophe of another Great Recession.

First appeared in Newsroom, April 8 2020

Simon Thornley: Do the consequences of this lockdown really match the threat?

As the Covid-19 picture emerges, it is vital to continually assess our response. The virus was identified quickly and tests developed. We are acquiring knowledge about it at a great rate. As cases mount across the world, a picture is also emerging of the effect of the virus on populations, which, as an epidemiologist, is my interest.

Unprecedented social controls have been rapidly thrust upon us. The justification initially was not overloading intensive care facilities, but we have now moved beyond that to “flushing out the cases we already have”. The duration of the lockdown is uncertain. It is also unclear how much of a financial hit the country is willing to stomach.

We know this virus is serious, but exactly how serious? How does the case-fatality rate, a measure of the importance of the disease, compare with other similar viruses?

As a rough guide, the US Centers for Disease Control and Prevention  uses rates of between 0.1 to 2.0 per cent to determine how to respond to a new threat. This is clearly an important statistic to attend to, but it is easy to get this wrong – very wrong.

The calculation is skewed, initially, because sicker patients are tested first, making the infection appear more serious. Also, in determining fatalities, uncertainty arises in patients who are otherwise sick with a limited life expectancy, who then test positive and subsequently die. Should they be labelled as Covid-19 deaths?

Remember swine flu in 2009? Initial estimates of case-fatality rates were about ten times higher than those calculated once the dust had settled. It turned out that swine flu, that year’s killer virus, was no more harmful than seasonal flu.

So let’s look at a couple of examples where more comprehensive testing has been completed. The Diamond Princess ship is one of the few examples of a closed population who were all tested for the disease. Seven deaths occurred in 700 test-positive patients, giving us a case-fatality rate of 1 per cent.

Remember, this was an elderly population. Calculations show that, if these rates were translated to a Western country’s overall age structure, the statistic would be 0.125 per cent (interval of plausible values: 0.025 per cent to 0.625 per cent), higher than normal flu (~0.1 per cent), but not by much.

The figures of up to 900 deaths a day in Italy are alarming, and so is the nation’s crude case-fatality rate of 9 per cent. However, a recent analysis of the deaths in Italy shows that only a small fraction were entirely due to Covid-19, occurring in people with no co-morbidities (3 out of 355; 0.8 per cent). Many deaths were hastily labelled as Covid-19 related when they were not.

As the average age of those dying is 80, that is not surprising. This kind of seasonal epidemic of deaths has occurred frequently in elderly populations living in this region for some years.

Another question to ask here is whether Covid-19 represents an added burden on top of usual seasonal viruses. After all, admissions to hospital, intensive care, and deaths occur at a background rate.

Time-series plots of overall deaths in European countries show surprisingly low rates for this time of year, even in heavily affected countries, such as Germany, Spain, France and Italy, even in the over-65 age group. Italy has the most dramatic increase, but no higher than occurred during the same season two years ago.

Is a “lockdown” and closing the borders even effective? Unfortunately, meta-analysis of social distancing measures for avoiding viral chest infections found that such an intervention was not strongly supported, since little evaluation of these policies had been done. Of all the preventive measures that were examined, improving hand hygiene had the best supporting evidence. It remains to be seen whether lockdown will result in “flattening the curve”, but we don’t have strong evidence in favour.

Despite my scepticism, Covid-19 does pose a real risk to our health. Sensible measures include better hand hygiene, ensuring good cough etiquette, and restricting large gatherings. Limiting exposure for the elderly, and people with chronic conditions, makes sense.

It is important that the public health response matches the threat posed to our health. It is important we keep abreast of developments, such as tests of immunity, so that we can return to normality quickly.

We don’t want to squash a flea with a sledgehammer and bring the house down. I believe that other countries, such as Sweden, are steering a more sensible course through this turbulent time.

Simon is a senior lecturer and epidemiologist at the University of Auckland.

Grant Schofield: Why we need a Plan B

Grant Schofield, Professor of Public Health at Auckland University says we need to exit lockdown.

Yesterday morning my teenage boys Sam and Jackson were out riding their bikes. The police pulled them over to question them about what they were doing and where they were going. They explained they were cycling for exercise and staying local. That seemed OK, but they then were told that they were sweating and sweating was dangerous, because it could transmit the virus, so they should avoid that. Wrong of course, but wow, unprecedented police powers that are beyond what New Zealanders have ever seen outside of wartime.

All in the name of eradicating Covid-19. Unite against Covid-19!

So can we achieve this?

The simple answer is no. We could with better tools. But now no.

Director-General of Health, Dr Ashley Bloomfield, recently said there is no plan B. The plan is the plan. In fact, there is strong social pressure to say nothing negative about the war effort. Negative is usually defined as anything off script, no matter how scientific. Look at my colleague Dr Simon Thornley in the last few days questioning the data being used to make our big decisions.

Back to eradication.

To eradicate this virus first, we need to know who has it, with certainty. At the most basic level that would mean that we can know if our elimination has worked. It also allows us to quarantine those who do have it, including all the asymptomatic, so they can’t unknowingly infect others. We would also be able to send those infected and now immune back out into the world.

To know these numbers we need a near-perfect test. One which correctly identifies all those who are infected, or not infected. Any false negatives, those who in reality are sick, but are tested and released as fine, would be particularly problematic. You’d also want to test on a scale with large numbers randomly across the population. That’s the only way you can confidently rule out the existence of the virus.

How does our testing go under this scientific scrutiny?

The best available test is Real Time PCR. A sequence of biochemistry allows for the detection of virus RNA. If you have these you almost certainly have Covid-19. The problem is that a negative test doesn’t mean you don’t have it. In fact, there are limited data on the accuracy (or sensitivity) of PCR for this virus. In China, a small number of patients with clinically compatible illness were tested using PCR and it correctly identified as many as 72 percent and as few as 32 percent and was highly variable in repeat tests. For similar viral testing sensitivity runs at 50-70 percent correct positives.

What is most important is that this means between 30-50 percent of those tested are released back into the community when they are in reality infected.

This is made worse because mostly we just test those with obvious symptoms. It means we won’t catch any of those without symptoms – those we most need to catch. It’s likely we have large numbers of asymptomatic people already in the community unintentionally infecting many others. Data from Iceland show that this might be as high as 50 percent of cases.

Second, we haven’t and don’t have the capacity to do the accurate level of massive population testing we need to. Simple as that.

Third, even if the above were both 100 percent, we cannot tell if you have already been infected and recovered. We need antibody testing to become available for this to be known. This hopefully comes to NZ soon. Eventually it will and we will really know the true numbers of infections. This means we can calculate the all-important “case fatality rate” which is the key number to decode how to react to a pandemic. We really must know how lethal this virus is, and for who, compared to something we already know like the seasonal flu.

To be clear, we don’t know this number with any certainty, and the modelling used to predict deaths, and therefore government responses, depends on this as yet uncertain number. The WHO’s original estimate was 3.4 percent, with the latest estimates looking closer to 0.3 percent and has usually come down further with more accurate measurement in other pandemics. The figures are still very uncertain. Seasonal flu has a case fatality rate of 0.1 percent.

I got most interested in this three weeks ago when I became sick with a sore throat, swollen glands, which eventually changed to a dry cough, difficulty breathing especially at night time, and then a lung infection. Repeated calls to the Healthline and my doctor resulted in vague advice not to worry and “drink more fluids”. Everyone in my house ended up with symptoms, all less severe than me. This really piqued my interest in how we were missing the boat. I had travelled on aircraft and mixed with people from overseas (at Ironman NZ in Taupo). Yet, I was outside of scope for being tested.

Did I, and my family, have Covid-19? I don’t know, and that’s the point.

We were originally on a plan to “flatten the curve”. This means reduce the speed of infection, also called mitigation. This seems prudent given we know we have a potentially lethal virus circulating in our community, with no easy way to accurately detect it. As well, there are probably large numbers of asymptomatic people spreading. Mitigation aims to lessen the potential high demands on the health system so we can give those who need it the best quality of care.

To me, it seems most likely that we cannot contain this because we can’t measure it.

Yet, somewhere in the last week the strategy changed to an eradication policy. Even if full eradication isn’t the goal, if it’s just “seriously flattening the curve” that requires much wider, more accurate surveillance and then quarantine than we have.

Our current strategy has so many questions, but very few answers.

What’s the exit strategy for this eradication policy? When do we go back to work, how do we decide with reliable data, and what then? What do we do if we get more infections again? Do we go back to lockdown? What if we are not successful?

Do we just “try harder” and lock down more? When do we decide to give up and try something else? How much harm is done to health from economic collapse and poverty will we get? How compliant will New Zealanders be in the longer-term? Even a small amount of civil disobedience is essentially unenforceable?

In the end, the goal in public health is to get the best possible outcomes, for the least possible harm, with the tools and resources we have available now. So as more and more data emerge, and we see where we are at we must be willing and able to change our mind, and move in ways that maximise the best health outcomes and the least harms.

Given we currently have no way of identifying accurately enough who has this virus, especially asymptomatic positives, then mitigation is the most logical pathway, especially given the unknown and potentially large harms of extended lockdown. Let’s explore our Plan B. Now is good.